Clinical Decision Support Systems

An Artificial Intelligence Medical Systems Prepared by : Dr.S.Lakshmi Pradha

DEFINITION “ A CDSS is defined as any software designed to directly aid in clinical decision making in which characteristics of individual patients are matched to a computerized knowledge base for the purpose of generating patient specific assessments or recommendations that are then presented to clinicians for consideration”.

FOUR KEY FUNCTIONS Administrative Managing clinical complexity and details Cost control Decision support

TYPES based on their USAGE Knowledge-based systems Alerts and reminders Diagnostic assistance Therapy critiquing and planning Prescribing decision support systems Information retrieval Image recognition and interpretation Expert laboratory information systems Machine learning systems

KNOWLEDGE BASED SYSTEMS Also known as expert systems contain clinical knowledge specifically defined task, and are able to give reasons with data from individual patients form of a set of rules

ALERTS AND REMINDERS In real-time situations Can warn the changes in patient’s condition Might scan laboratory test results, drug or test order, or the EMR. Send reminders or warnings, either via immediate on-screen feedback or through a messaging system like e-mail. Reminds to notify clinician of his important tasks .

DIAGNOSTIC ASSISTANCE Help in the formulation of likely diagnosis, based on patient data presented to systems, and its understanding of illness, and storage of knowledge base With complex data, such as the ECG Prevent missing of rare clinical presentations of common illnesses like myocardial infarction Prevent struggles relating to formulating diagnosis

THERAPY CRITIQUING AND PLANNING Look for inconsistencies, errors and omissions in an existing treatment plan Do not assist in the generation of the treatment plan Warn or guide during physician order entry

PRESCRIBING DECISION SUPPORT SYSTEMS: PDSS Checking for drug-drug interactions, dosage errors, and if connected to an EMR, for other prescribing contraindications such as allergy They support a pre-existing routine task Improving the quality of the clinical decision Automated script generation and electronic transmission of the script to a pharmacy.

INFORMATION RETRIEVAL Can assist in formulating appropriately specific and accurate clinical questions Act as information filters Assist in identifying the most appropriate sources of evidence to a clinical question More complex software ‘agents’ can be used to search and retrieve information to answer clinical questions contain knowledge about its user’s preferences and needs have some clinical knowledge to assist it in assessing the importance and utility of what it finds

IMAGE RECOGNITION AND INTERPRETATION Automatic interpretation of plain X-rays and more complex images like angiograms, CT and MRI scans This is of value in mass-screenings

EXPERT LABORATORY INFORMATION SYSTEMS Whole report of a patient is generated by a computer system that has automatically interpreted the test results Do not intrude into clinical practice Embedded within the process of care, and allows clinicians to concentrate in patients, also does not expect the laboratory staff or the clinicians to interact On clinician ordering, system prints a report with a diagnostic hypothesis for consideration, but cannot be made responsible for information gathering, examination, assessment and treatment Thus the system cuts down the workload of generating reports, without removing the need to check and correct reports

MACHINE LEARNING SYSTEMS One of the driving ambitions of Artificial Intelligence has been to develop computers that can learn from experience They attempt to discover humanly understandable concepts Learning techniques include neural networks learn from decision trees with examples taken from data

MACHINE LEARNING SYSTEMS continued Can produce a systematic description of those clinical features that uniquely characterise the clinical conditions Knowledge in the form of simple rules or as a decision tree Classic example is KARDIO , which was developed to interpret ECGs

MACHINE LEARNING SYSTEMS continues Can be extended to explore poorly understood areas of healthcare Process of ‘ data mining ’ and of ‘ knowledge discovery ’ systems Learning system For example : takes real-time patient data obtained during cardiac bypass surgery, and then creates model of normal and abnormal cardiac physiology These models might be used to look for changes in a patient’s condition Used in a research setting: these models can serve as initial hypotheses that can drive further experimentation

MACHINE LEARNING SYSTEMS continued One particularly exciting development : to discover new drugs . Based upon the new characterisation of chemical structure produced by the learning system, drug designers can try to design a new compound that has those characteristics. Less time in development of new drugs and the costs is significantly reduced.

MACHINE LEARNING SYSTEMS continued Machine learning has a potential role to play in the development of clinical guidelines For example in a case that there are several alternate treatments for a given condition, with slightly different outcomes. Machine learning systems can be used to identify features that are responsible for different outcomes.

BENEFITS Improvement in Patient Safety Reduction in Medication Errors and Adverse Drug Events Enhancement of Prescribing Behaviour Improved Quality of Care Improved Compliance with Clinical Pathways and Guidelines Time Release for Patient Care Improved Efficiency of Health Care Delivery Processes

WHY MORE CDSS ARE NOT IN ROUTINE USE Some require the existence of an electronic patient record system to supply their data. Others suffer from poor human interface design and so do not get used even if they are of benefit. Required additional effort for already busy individuals. The technophobia or computer illiteracy of healthcare workers. If a system is perceived by those using it to be beneficial, then it will be used. If not, independent of its true value, it will probably be rejected.

REASONS FOR SLOW IMPLEMENTATION Lack of formal evaluation of these systems, Challenges in developing standard representations, Lack of studies about the decision making process, Cost Difficulties involving the generation of knowledge bases, Practitioner skepticism about the value and feasibility of decision support systems.

LIMITATIONS Some CDSSs follow the costly development path of medical devices and FDA approval. The majority of systems are not bound by these vigorous criteria. Generally, CDSSs are proliferating as fragmented and isolated systems in a few clinic- or hospital-wide exceptions in academic centers. (In parallel, the public awareness of safety and quality has accelerated the adoption of generic knowledge-based CDSSs ). Another barrier, the structured data entry process, remains a challenge for all clinical information systems including CDSSs .

LIMITATIONS OF EVALUATION COMPONENTS OF CDSS:- A focus on post-system implementation evaluation of users’ perceptions of systems. Rely upon the retrospective designs which are limited in their ability to determine the extent to which improvements in outcome and process indicators may be causally linked to the CDSS . Rare adoption of a comprehensive approach to evaluation where a multi-method design is used to capture the impact of CDSS on multiple dimensions.

LIMITATIONS OF EVALUATION COMPONENTS OF CDSS:- Concentration on assessment of technical and functional issues. Such evaluations have also failed to determine why useful and useable systems are often unsuccessful. Expectations that improvement will be immediate. In the short term there is likely to be a decrease in productivity. Implementing information systems takes time and measuring its impact is complex thus a long-term evaluation strategy is required. Almost none use naturalistic design in routine clinical settings with real patients and most studies involved doctors and excluded other clinical or managerial staff.

DISADVANTAGES Substantial cost of knowledge acquisition and knowledge maintenance Lack of rigorous studies (i.e., clinical trials) to identify evidence that supports CDSSs Lacks clarity of legal and economical implications of sharing such knowledge bases

FOUR BASIC COMPONENTS Inference engine (IE) Knowledge base (KB) Explanation module Working memory

INFERENCE ENGINE (IE) Main part of the system Controls on what kind of action taken by the system, by using the knowledge in the system and knowledge about the patient to draw conclusions on specific conditions. Determines the route of alerts and reminders in an alerting system or conclusions to be displayed in a diagnostic system.

KNOWLEDGE BASE (KB) Knowledge used by the IE Built up with the help of a domain expert or by a automated process. A knowledge engineer with the help of clinical domain expert creates, edits and maintains KB. In an automated process, knowledge is acquired from external resources such as databases, books, and journal articles by a computer application. Example :Protégé

WORKING MEMORY Collection of Patient data may be stored in a database or as a message Demographics Allergies Medications in use Previous dental and medical problems Other information

EXPLANATION MODULE For composing justifications for the conclusions drawn by the IE in applying the knowledge in the KB against patient data in the working memory This component may not be present in all the CDSS .

FUNCTIONALLY CLASSIFIED AS Synchronous mode Asynchronous mode Open loop system Closed loop system

FUNCTIONALLY Synchronous mode CDSS communicates directly with the user who is waiting for the output of the system. For ex : checks for drug-drug interaction when the provider is writing a prescription Asynchronous mode CDSS performs their reasoning independently of any user awaiting its output. For ex : generation of a reminder for an annual visit for check up and hygiene.

FUNCTIONALLY In open loop system, CDSS draws the conclusions but takes no decision directly of its own. For Ex : An application that generates an alert or a reminder. The final decision is taken by the clinician. In a closed loop, the action can be implemented directly without the intervention of the human.

OTHER TYPES:- Event monitors Consultation systems Clinical guidelines An Event monitor is a software application that receives copies of all data available in an electronic format in an institution and uses its knowledge base to send alerts and reminders to clinicians when deemed appropriate.

In Consultation systems, a clinician enters details of a case (patient demographics, complaint, physical examination findings, test results etc) into the system, and the system in turn provides a list of problems that may explain the case and suggests actions to be taken.

Clinical guidelines are incorporated into the CDSS . They are developed by group of experts and disseminated by the government or by professional organizations. They represent formal statements of recommended best practices with regard to a particular health condition. To improve sharing of such guidelines, researches have tried to develop standard knowledge representations such as Arden Syntax or Guide Line Interchange Format (GLIF).

Arden Syntax Is an ANSI standard representing commutable clinical knowledge. Each decision rule is called Medical Logic Model (MLM). Each MLM has sufficient logic to make single clinical decision

Arden Syntax For Example :- evoke : /* evoke on storage of a serum digoxin level */ storage_of_digoxin;; logic : /* exit if the digoxin level is 0 */ if digoxin <= 0 then conclude false; endif; /* get the last valid potassium */ potassium := last(raw_potassiums); /* exit if no hypokalemia is found */ if potassium < 3.3 then ; /* send an alert */ conclude true; else conclude false; endif; ;;

Arden Syntax action: write “The patient’s serum digoxin level indicates that the patient is taking digoxin. The patient’s most recent potassium level is low, and the hypokalemia may potentiate the development of digoxin related arrhythmias.”; ;;

Guideline Interchange format A computer-interpretable format for the representation of clinical practice guidelines developed by the InterMed collaboration (a joint project of laboratories at Harvard, Stanford, Columbia,and McGill universities). Designed as a general purpose language for development and implementation of guideline-based clinical decision support systems with applications in different clinical domains. Provides patient-specific recommendations. Used for quality assurance and medical education.

BASED UPON KNOWLEDGE REPRESENTATIONS CDSS can be classified into Algorithmic Neural networks Probabilistic Logical/deductive (rule-based) Hybrid systems

ALGORITHMIC SYSTEMS Use logical classification methods, represented as decision trees and flowcharts that lead the user to a desired end point. Does not depend on large sample sizes of data and can be applied across patient populations. DISADV Lack of flexibility with which the decision points are incorporated into the statements of the program. This method does not incorporate uncertainty. Changes in knowledge may require substantial rewriting of the system. In a complex system, decisions may be impossible to understand and revise. For Example : - Recommendation of chemotherapy drugs for breast cancer and a diagnostic aid for oral pathology.

NEURAL NETWORKS (NN) Are algorithms that require training to create a set of solutions to a problem. After training, these algorithms can make decisions on new problems with incomplete facts; they are commonly used in pattern recognition problems. first implemented as a biological model of the brain in 1940s successful at narrow and well-defined clinical problems such as classifying textual output of images, diagnosis support, Prognosis evaluation. Commercialized for image recognition and are used in Uterus cervix cytology labs. in dentistry to identify people at risk of oral cancer and pre-cancer for lower third molar treatment planning decisions.

PROBABILISTIC SYSTEMS Incorporate rates of diseases or problems in a population and the likelihood of various clinical findings in order to calculate the most likely explanation for a particular clinical case. Typically employ Bayes Theorem , which is a mathematical model that accounts for the prevalence of disease in a population and the characteristics of a particular patient to calculate the probability that a particular patient has a particular disease. Advantage Their output reflects the relative likelihood of diagnosis or success of treatment, they may be limited by the fact that the necessary probabilities either are not known or are derived from a population at least somewhat different from the patient in a particular case. Examples of Bayesian systems in dentistry :- The Oral Radiographic Differential Diagnosis (ORAD)

LOGICAL/DEDUCTIVE SYSTEMS A collection of “ if-then ” rules—to make decisions While the “ if-then ” rules of a logical/deductive system allow representation of the branching questions used by experts to make clinical decisions, they may overemphasize certain diseases if they are not adjusted for the rarity or prevalence of particular diseases. Examples Bleich’s software that diagnosed acid-base disorders Application in dentistry is RHINOS , a consultation system for diagnosis of headache and orofacial pain.

CRITIQUING MODEL Reacts to proposed diagnosis or treatment with agreement or alternatives Examples of such a system are HT-ATTENDING, HyperCritic, RaPiD Both HT-ATTENDING and Hypercritic are systems designed to critique the management of hypertensive patients. RaPiD uses both an automated and critiquing model for removable partial denture design.

HYBRID SYSTEMS AIM : to overcome these drawbacks by combining both deductive rules and probabilistic reasoning in the same CDSS They use features of several or all the previously described systems along with heuristics to assist clinicians in making decisions. Example: HEME , a system used to diagnose blood diseases

FUTURE On the adoption of evidence-based practice , Progress in developing useful programs, Adoption of standards to allow interoperability, Reduction of logistical barriers to implementation, Understanding of the complex and changing nature of clinical knowledge, and proper validation of the programs. Overcoming challenges related to the legal implications inherent to the development and use of such innovations.

THANK YOU Prepared By Dr. S. Lakshmi Pradha

Clinical Decision Support Systems

In this document

More Related Content

What's hot

Similar to Clinical Decision Support Systems

Clinical Decision Support Systems